consensus-publish-guard

What This Skill Does

The consensus-publish-guard acts as a final, high-assurance safety gate for all public-facing outbound content. Whether you are automating blog posts, scheduling social media updates, or drafting technical announcements, this skill ensures that your message aligns with established organizational policies and stakeholder expectations. By utilizing persona-weighted governance, it evaluates drafts not just through a single lens, but through a collective decision-making framework. The guard returns one of three critical status codes: APPROVE, BLOCK, or REWRITE. If a rewrite is triggered, the skill provides a targeted patch to bring the content into compliance. All final decisions and the reasoning behind them are persisted as immutable audit artifacts within your board state, ensuring full transparency and accountability.

Installation

To install the skill, use the ClawHub CLI command: clawhub install openclaw/skills/skills/kaicianflone/consensus-publish-guard. Because this skill relies on the consensus-guard-core dependency, you must ensure your development environment is set up with a local sibling directory structure. Navigate to your repository root and ensure the core package is accessible before running npm install. The skill reads environment variables like CONSENSUS_STATE_FILE to manage its persistence layer, so ensure these are configured in your local environment file.

Use Cases

This skill is designed for scenarios where brand reputation and legal accuracy are paramount. Primary use cases include:

• Automated Social Media: Validating post tone and factual claims before they reach public feeds.
• Product Marketing: Ensuring feature claims in product launches match internal product specs and legal disclaimers.
• Policy-sensitive Communication: Screening internal-to-external communication for data exposure or tone violations in highly regulated industries.

Example Prompts

1. "Guard this blog post draft: check against 'marketing-persona-set-01' and provide a rewrite if the claims about Q4 earnings are not sufficiently caveated."
2. "Review the current output for the social campaign using mode=external_agent with the votes provided in the attached metadata."
3. "Run consensus-publish-guard on the pending release notes; if blocked, generate the artifact explaining which policy check failed."

Tips & Limitations

To maximize the utility of the guard, ensure your persona_set definitions are granular and frequently updated to reflect current organizational goals. The tool is highly deterministic, meaning it will consistently block or allow content based on the policy file provided; it does not 'guess' intentions but follows explicit rule sets. A key limitation is that this skill does not perform external network calls, meaning it cannot verify live, real-time data from the web—it relies entirely on the state stored in your board artifacts. Always ensure your consensus-guard-core is kept updated to the latest pinned version to maintain compatibility with updated board primitives.